In week 49, #MakeoverMonday looked at an article published on FT.com which analysed the cost of eating out at JD Wetherspoons, a popular pub chain in the UK. There are over 900 branches of Wetherspoons in the UK but for this analysis the FT had calculated the cost of exactly the same meal at 213 different branches using the Wetherspoons app. This was the meal they ordered:
They then analysed the price differences between pubs using a variety of charts, including the two below:
What do I like about the original?
- The analysis is featured within a comprehensive article which fully supports each visualisation provided.
- There is more than one chart featured within the article, each providing a different prospective on the data.
- The charts are simple and can be understood quickly.
What could be improved?
- The charts aren’t interactive. If you’re trying to identify a particular Wetherspoons branch you may struggle.
- The visualisations aren’t the most attention-grabbing. The line chart for instance is very basic.
My approach to the dataset
I wanted to try something new this week. I have suffered with a bit of a creative block recently and have struggled to find ideas for my visualisations. This isn’t like me; usually I am thinking of ideas for visualisations all day long! While I have been working on #MakeoverMonday each week I haven’t published all of my visualisations yet because I’m simply not happy with them. I hope to revisit them over the next few weeks and rework my designs until I’m happy to publish.
I initially explored the data and built some basic visualisations to familiarize myself with the content. Nothing was really standing out to me that wasn’t already obvious; pubs in London are more expensive and pubs in less desirable areas of Birmingham are cheaper. Initially I intended to try and prove or disprove the theory that pubs in University cities or towns are cheaper. However, I decided against this in the end as I didn’t feel the analysis would be very interesting. As a side note I found a pub I used to visit frequently in my student days in the data; The Charlie Hall in Erdington, Birmingham. This is certainly not a pub I wish to visit again anytime soon! When I used to visit back in 2003/4, smoking indoors in the UK was still legal. Upon entering this pub you were greeted by a strong stentch of smoke from the smoking area at the front. The pub was dark and dingy and during the day used to be full of elderly men; some of which I’m certain would sit at the bar from morning until closing time. On one visit we even witnessed an older gentleman fall off his chair after consuming a few too many pints! Anyway, I digress…
Going round in circles
There seem to be a lot of radial charts being produced at the moment. These charts are usually way out of my comfort zone! I admire the work of authors such a Rody Zakovitz and Neil Richards who are fans of radial charts but I wouldn’t have a clue where to begin building such a chart myself.
A recent visualization by Alexander Waleczek (Axel) caught my eye a few months ago. This visualisation was a #MakeoverMonday looking at the number of public holidays in different countries around the world:
(click on the picture to interact with the visualisation on Tableau Public)
Axel’s viz was later featured on Tableau Public as both Viz of the Day and Viz of the Week. Inspired by this chart I knew I wanted to produce something similar. The problem was I didn’t know what this chart type was called! I initially referred to Axel’s blog post explaining his design process but unfortunately he did not name the chart type in his blog. Next I googled ‘Sunburst Chart’ as I thought this might be the correct name. Unfortunately this returned a gallery of charts like those in the screenshot below which wasn’t what I was looking for.
Next I contacted Neil Richards for help. Surely he would know the name of Axel’s chart?! Unfortunately Neil also thought these were called Sunburst Charts and like me quickly realised this was not the case. Neil suggested I look at Rody’s or Axel’s blog but unfortunately I had already checked both with no success. While Rody has published charts like these before I couldn’t find a tutorial on his blog which explained them.
I began googling under different names until I eventually found a blog post by Ranjeev Pandey which described the chart I wanted to build; a Radial Bar Chart! Ranjeev’s blog was inspired by an older blog post by Dave Hart from Interworks. I struggled to find this article because the link was broken everywhere I saw it referenced. Thankfully I managed to find the original eventually and ended up following this post to build my chart. The blog post was surprising easy to follow and I built the radial bar chart I wanted fairly quickly. If I can do it, anybody can!
For this visualisation I took inspiration from the JD Wetherspoons website. The Wetherspoons colour palette primarily consists of black, white and gold/yellow so I stuck with these colours. I wanted to colour code the bars by colour so I split the meal prices into 3 buckets. I wanted my visualisation to focus on the most expensive and the cheapest places for a meal; I wasn’t too concerned about those where the prices were ‘average’. With this in mind I coloured those in the middle price bracket light grey so attention would be drawn to the black (the most expensive) and the yellow (the cheapest) pubs.
Wetherspoons also use an informal, hand-writing style font on their website. I wanted to use something similar in my visualisation. While I couldn’t find the exact one I decided upon a free font called “KG Ten Thousand Reasons” by Kimberly Geswin which I found on this website. Obviously this is a non-standard font and is therefore not supported by Tableau Public. To get around this I wrote all the text in Word and took individual PNG images of each piece of text I needed. I then added these to my dashboard as images.
To add further interest I added some call-outs for the most/least expensive pubs, including photos of each one. I was hoping to include more boxes but I ran out of space therefore limited myself to 3.
Here is my final visualisation:
Thanks for reading.